Time on page number seems massively incorrect for some pages and is inflated

stephan1994w commented 2 years ago

Past Issues Searched

[X] I have searched open and closed issues to make sure that the bug has not yet been reported

Issue is a Bug Report

[X] This is a bug report and not a feature request, nor asking for self-hosted support

Using official Plausible Cloud hosting or self-hosting?

Plausible Cloud from plausible.io

Describe the bug

We have an article page that is registering insane Time on Page values. Yesterday the average time on page was over 5 hours. In the documentation it says this should be the average time spent by users, but this seems incorrect. Any support here would be greatly appreciated

Expected behavior

Expected value of Time on Page to be substantially lower. It's roughly 800 page views, to get this kind of value, it implies there are people spending long periods of time on the page and this cannot be correct as the page content does not justify this.

Screenshots

Environment

- OS:MacOS
- Browser: Chrome
- Browser Version: 104.0.5112.101

metmarkosaric commented 2 years ago

thanks for reporting @stephan1994w! time on page does show the average time people spend on a particular page on your site.

we have a different "time on page" definition than GA which is where some differences come from. In GA, bounced visitors are included in the time on page calculation but are set to say 0 or 10 seconds. in Plausible, we completely exclude bounced visitors from the time on page calculation. so say you have a very high bounce rate on a specific page, we would not count all those bounced visitors for time on page so those few that do stick around for a long time will contribute to a very high time on page.

neither of these approaches is 100% accurate as it's not possible to measure the time on page of people that bounce from your site. Plausible would typically show higher than the actual time on page while GA would typically show lower than the actual time on page.

we're working on a new approach with the beacon API, which would allow us to include actual time on page even from people that bounce, making Plausible have more accurate time on page than what's possible now in web analytics but for now we continue the "flawed" method like other analytics tool

jkettmann commented 1 year ago

I'm also curious as I see very long durations as well. A few questions:

What happens when a user just leaves the tab open and thus never leaves the page (or only after days or weeks)?
What is a bounced visitor in this context? My bounce rates are very low because I have a custom event that triggers for almost every visitor. So I understand that these are not considered when calculating the bounce rate, right? But are they considered as bounced for the calculation of time on page?

dnl4 commented 2 months ago

What is a bounced visitor in this context? My bounce rates are very low because I have a custom event that triggers for almost every visitor. So I understand that these are not considered when calculating the bounce rate, right? But are they considered as bounced for the calculation of time on page?

Hi @jkettmann, 2 years later i've the same question as you.

I've created a new page that only I have access to, this page has a custom event the trigger every 5 seconds, for test purposes. But the "time on page" keep showing 0 seconds.

Could @metmarkosaric kindly help me understand this, please?

metmarkosaric commented 2 months ago

hi @dnl4! "time on page" is calculated as the difference between the point when a person lands on a particular page and when they move on to the next page. custom events do not impact the time on page calculation (they do impact bounce rate and visit duration calculations). see all definitions here.

plausible / analytics